Junhan Wang1 ans Jonathan Sahagun2, 1USA, 2California State Polytechnic University, USA
In this paper, we tackle the pressing communication gap between the Deaf and hearing communities, an issue affecting millions of individuals worldwide [1][2]. The proposed solution is a machine learning-powered application that translates sign language into text in real-time, allowing Deaf and hearing individuals to communicate directly [3]. The development faced challenges such as acquiring a diverse and accurate dataset and managing real-time processing of gestures. Experimentation involved testing the model's accuracy across multiple users, revealing promising outcomes. Two existing methodologies, a sensor-based glove and a single-camera solution, were compared, highlighting areas for potential enhancement in our approach. Despite the diversity of sign languages and their unique grammatical structures, the project represents a significant step towards more accessible communication. It highlights the potential for further advancement in machine learning applications for translation and inclusivity.
